Plot multiple series in the secondary y-axis in Gnuplot
by MARCO TEDESCHI
Dear all,
I would like to ask whether it is possible, within the gnuplot command in Gretl, to assign multiple series to the secondary y-axis. In the example script below, the aim is to plot both heating and gasoline prices on the y2 axis. I tried using a list of variables in the y2 option, but it does not work as expected. Is there a way to specify more than one series for y2 using the gnuplot command? Any suggestions or workarounds would be greatly appreciated.
Best regards,
Marco
EXAMPLE SCRIPT:
include yahoo_get.gfn
nulldata 1500
setobs 5 2018-01-01
series wti = yahoo_price("CL=F",1)
series brent = yahoo_price("BZ=F",1)
series gasoline = yahoo_price("RB=F",1)
series heating = yahoo_price("HO=F",1)
delete index
list Energy = wti brent gasoline heating
list y2 = gasoline heating
gnuplot Energy --time-series --with-lines --output=display --y2axis=gasoline
2 months, 3 weeks
random number generation in gretl
by Cottrell, Allin
This is a heads-up about a recent change in gretl git, now in the
latest snapshots at sourceforge. We decided to replace the Mersenne
Twister (SFMT) as gretl's pseudo-random number generator with
xoshiro256+, a modern, lightweight and cleanly coded RNG with very
good statistical properties. You may find it a little faster than
SFMT.
One point to note: if you have gretl scripts that use "set seed" in
order to produce a replicable sequence of pseudo-random variates, the
sequence will now differ with the new RNG. We had hoped to maintain
use of SFMT as an option but it turned out that was just too
complicated.
Further reading for anyone who's interested:
"A PRNG shootout": https://prng.di.unimi.it/
"It is high time we let go of the Mersenne Twister":
https://arxiv.org/abs/1910.06437
Allin Cottrell
2 months, 3 weeks
fRUSTRATION TRYING TO LOAD AN XLSX DATASET
by Brian Revell
Hi
I have an xlsx datafile comprising 10 rivers labelled 1..10, 24
observations dated 2000-2023 plus 5 variables (See attached screenshot). I
have use File open command specifying it is an xlsx file which it reads and
then produces an error message.
F*ound 1 valid sheet*
*Found 11 Variables and 240 observations!!!*
*Name missing for variable 7 !!!*
I am using Gretll version 2025c
I get the same error message if I specify File, New Data set, Panel data
with n=10 and t=24 and then append the data.
The full dataset is also appended for information. Advice welcomed as I
cannot move forward without getting Gretl to read swhat is a basic XLSX
file. I have in earlier versions had no problems loading
similararly structured stacked time series for panel data analysis
*Brian Revell*
2 months, 3 weeks
Which libopenblas library under Debian/ Ubuntu?
by atecon＠cryptolab.net
Hi all,
I am planning to install Gretl on a heavy machine with many cores (=96). In the Ubuntu repo there exist 4 different libopenblas packages:
0) libopenblas0
1) libopenblas0-openmp
2) libopenblas0-pthread
3) libopenblas0-serial
For each package another 4 corresponding 64-bit versions (shared lib, 64bit) exist, too.
Which libs would you recommend on:
1) a standard private laptop with modern Intel oder AMD CPU
2) a "cluster" with many CPUs (AMD tech. actually) and even GPUs?
I could not find anything useful about this, except Marcin's paper mentioning "libopen-blas0-openmp 0.3.20+ds-1":
https://link.springer.com/article/10.1007/s00180-024-01461-w
Thanks
Artur
2 months, 3 weeks
gretl4py 0.4 released
by Marcin Błażejowski
Hi,
I've just released gretl4py v0.4, which is available via PyPi
infrastructure for:
- ms windows, both: x86-64 and arm64
- macOS (signed), both: x86-64 and arm64
- linux, x86-64 only
Note: macOS users shall install the package with "--user" option.
What's new?
- apparatus for slicing datasets (via '[]' operator)
- GretlaDataset methods: ::dummyfi(), ::head(), ::tail(),
::to_dataframe(), ::lags(), ::square(), ::discrete(), ::logs(),
::orthdev(), ::stdize()
- '--unit' and '--time' options to GretlaDataset::sample()
- Pooled OLS tweaks
- add 'n' and 'k' properties do GretlDataset
- add 'range' option for GretlDataset::print()
- add nls, mle, and gmm estimators
- add system estimation (all methods available in gretl)
In addition, some bugs fixed, massive code refactoring, and revert to
gcc-based libgretl for windows amd64 builds due to some OpenBLAS issues.
Have fun!!!
Marcin
2 months, 3 weeks
Removing one Panel member dataset in Panel data analysis.
by Brian Revell
I am unable to figure out how to exclude one specific panel from a stacked
timeseries panel dataset of 5 members. Were that particular panel no.1 or
no. 5, it is straightforward. But this one is no.3
Is there a specific boolean expression to exclude that with ID=3? Simply
specifying ID = 3 or ID==3 does not seem to work.
Suggestions welcomed.
The other alternative is simply to remove Panel member data no.3 from the
xlsx file and reload into Gretl.
*Brian Revell*
3 months, 1 week
